Experimental and computational investigation of combustion phenomena in mesoscale ducts

Experimental and computational investigations of the flame phenomenology in mesoscale tubes were performed. Particular emphasis was given to oscillating flames, a phenomenon in which a flame front undergoes a periodic cycle of upstream propagation, extinction, and downstream re-ignition. Fuel-rich methane-oxygen mixtures and straight quartz ducts with one end open to the atmosphere were used. The flame behavior was determined as a function of equivalence ratio and Reynolds number in order to produce maps of the phenomenology and identify the boundaries between behavioral regimes. Rich propane-oxygen mixtures and multiple tube lengths were also studied in order to examine the impact of fuel selection and geometry on the regime boundaries. Infrared thermometry was utilized in order to determine the wall temperature distributions for different flame phenomenologies and to test the hypothesis that the moving nature of an oscillating flame will more uniformly distribute enthalpy throughout the duct. Both quartz and steel tubes were studied during IR experiments in order to investigate the effects of material properties on the resulting temperature distribution and flame phenomenology. Analysis of the exhaust gases was performed in order to measure fuel leakage and incomplete combustion that may occur due to the gaps in combustion that exist in the oscillation cycle. Oscillating flames were found to have fuel leakage and pollutant emission no greater than conventional stationary flames, and to produce a nearly uniform temperature in the tube wall, aspects that make them highly suitable for small-scale power generation. The results of the experiments were used in efforts to develop a high-level computational model that could predict the flame phenomenology. The experimental results strongly suggested a thermal driving mechanism underlying oscillations and revealed the interwoven nature of the wall temperature distribution and the flame phenomenology. The computational model succeeded in qualitatively reproducing all observed flame phenomenologies. However, it was also determined that more complex mechanisms are needed in order to accurately replicate the experimentally observed boundaries between behavioral regimes